It depends on how you have the timer thread/service/whatever deployed. If you
have one running for every server in the cluster, then yes, it will be an issue.
What we did, is only have ONE server run the timer thread/service, so they only
get run from one source to ensure there is no cluster
fmarchioni wrote :
| Better finding a SERIOUS company..
| Good luck !
|
Like what Ronald mentioned... this is an open-source project, and this forum is
maintained by developers/users of jBPM, not paid JBoss employees.
We see way too many posts where people want to know how to do the
You need the c3p0-0.9.0.jar (Assuming hibernate 3.1.x) in your classpath
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4002290#4002290
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=4002290
The path environment variable is not the same as your classpath.
Readup on java and what the classpath is, and how to set it up in your
environment. Then add the .jar to the classpath.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=4002393#4002393
Reply to
The java tutorial is always a good start for new java developers
http://java.sun.com/docs/books/tutorial/
Then, learn eclipse and how to define projects/classpaths/etc. Either check
eclipse.org or poke around until you figure it out. Or ask one of the java
resources you have there.
View the
You need to make sure the scheduler is available then.
In your jbpm.cfg.xml, make sure you have service name=scheduler
factory=org.jbpm.scheduler.db.DbSchedulerServiceFactory /
Inside the jbpm-context element.
This should have been an easy find on the forum, I believe it's been asked many
If you have searched, with no luck, then at least you tried, I can't get too
upset. I just get irritated with the number of posts on this board from people
who obviously didn't try to figure it out on their own, so I apologize for
being short.
Anyway, looking at your web.xml, the servlet
Thanks Ronald. I couldn't really 'validate' the web.xml that was there, but
just glancing at it, it 'looked' right.
We programmatically start our use of jbpm's Scheduler service, which is always
an option as well.
View the original post :
SchedulerSession.findTimersByDueDate is a named query inside
hibernate.queries.hbm.xml
It is important that the hbm.xml taht jBPM uses has the following in it:
mapping resource=org/jbpm/db/hibernate.queries.hbm.xml /
View the original post :
I'm sorry, I meant that the hibernate.cfg.xml needs to have that mapping line
in it
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3999484#3999484
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=3999484
your hibernate.cfg.xml needs to have all the mappings for jbpm.
Look inside the jbpm-3.1.2\src\config.files directory, that hibernate.cfg.xml
needs to be referenced by jBPM. ALl the mappings defined in that cfg.xml need
to be in your system.
The jBPM JAR should have that file already inside
We actually have two separate hibernate.cfg.xml (the other is called
jbpm.hibernate.cfg.xml) You could just jam the contents of them both together
into one, or load up your SessionFactory/Configuration programmatically to read
both config files.
Either way you need the jbpm mappings.
View
Well, your cache isn't enabled. Make sure you have ehcache.jar in your
classpath/etc. (Or just disable second-level cache)
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3999556#3999556
Reply to the post :
jBPM already includes it's own .hbm.xml files.
Its probably trying to rollback when it fails getting the timer query, and then
bombs in rollback.
Search this forum, and the jBPM user guide to make sure your environment is
setup properly.
View the original post :
I am using a custom AssignmentHandler, and we have the identity component
removed.
We just extend from AssignmentHandler and never touch that
ExpressionAssignmentHandler.
Then just specify your new class as the appropriate handler, and you should be
set
View the original post :
You have to create the jbpm context first with
JbpmConfiguration.getInstance().createJbpmContext();
before you can get the current. Once you have created it, then you can get the
current with JbpmContext.getCurrentJbpmContext()
You've got to create something before you can get the current
You can always just do a hibernate query against the DB
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3992298#3992298
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=3992298
___
I think it is honestly how *you* want it to work.
Both scenarios you mention (never-ending or ending to be later restarted) are
easily supported, and just depends with what code you want to put in place in
front of jBPM.
Even if you support your 2nd scenario, anyone designing a flow could
Yeah, I've been burned by this recently too.
However, if you do a .toString() on the object, it displays the proper class
name with the name of the node in parens: 'EndState(End)'
My ghetto fix around this was to .toString() and then parse the text until the
first parenthesis.
The to string
Thanks, I'll give that a shot. I figured it was something like that, but I
didn't have the time to figure out a real solution (eep). Though my fix is a
hack, it works! ;)
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3986541#3986541
Reply to the post :
Olivier, I used your method, and it works like a charm, thanks.
I'm now using the instanceof like a proper developer should ;)
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3986593#3986593
Reply to the post :
It is easy to add action handlers to each event. Process and task
create/completion have their own events, and you can add your own custom
handlers.
Check out the user guide about actions, events, and Custom action example
View the original post :
Olivier_Debels wrote : AFAIK you don't need the link task/tasknode. You can
just have a task def tied to the process definition and create task instances
in the task node (using event handlers f.e.). You can actually create task
instances in any node.
We're already doing this. We have
Olivier_Debels wrote :
| Well I guess you can create your own custom task instance which can deal
with this. It just requires overriding of the end() function and keeping a
TaskNode reference. Haven't tried it but I guess this would work.
I don't think that would change my issue of
[EMAIL PROTECTED] wrote : didn't read the whole thread. sorry if i
duplicate an answer.
|
| ...but did you consider defining the tasks in the task-node and specify
create-tasks=false on the task node ? that way, can still fetch the task
definitions and create instances at runtime and
To add additional tasks during runtime, you'd have to create new task
definitions. These tasks are already defined and created, we just don't want
instances automatically started for every task def when the task node enters.
View the original post :
anonymous wrote : Then you can create a task instance of the mandatory task in
the task node
There's no way to tie a task instance to a task node. It's only through the
task def that there's a link between task/tasknode.
I'm already creating the other task instances using event handlers, but
You could always design the flow to have transitions to go backwards.
Or 'digest' the flow when they give you the .par file to process and
programmatically create 'reverse' transitions between nodes
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3976671#3976671
(I'm using jBPM 3.1.2)
We have task nodes, and are creating a number of tasks on each task node.
There will always be one mandatory task per task node, and then a number of
optional tasks per task node.
We would only like to create the optional task instances when the user wants,
not
Interesting, I ran into a similar, but still different issue.
http://jira.jboss.com/jira/browse/JBPM-680
I saw that one of the tests was creating the scheduler thread w/ the default
constructor, and:
anonymous wrote : Changing lines 146 and 185 from:
| new
jBPM is already released as a JAR file. You just include it (and any libraries
it depends on) in your application, and call the code like you would any other
library.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3974869#3974869
Reply to the post :
You've got to give more detail than I get into trouble
What errors are you seeing?
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3970990#3970990
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=3970990
Because you are using the JbpmContextFilter, which opens and closes the context
for you in your servlet. Your servlet should not create a context, but should
get the current context.
JbpmConfiguration jc = JbpmConfiguration.getInstance();JbpmContext jbpmContext
= jc.getCurrentJbpmContext();
You have all of the required Jars in your EAR.
Do you have all of the proper manifest.mf classpath entries?
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3969459#3969459
Reply to the post :
http://wiki.jboss.org/wiki/Wiki.jsp?page=JbpmOnTomcat
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3968962#3968962
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=3968962
___
I am pretty sure (but not 100%) that it drops/recreates the way it is right
now, because the JUnit tests rely on it to create a fresh schema each time.
But like Ronald said, looking at the code will answer everything :) That, or
just even running it to see if it kills your database (try it on
That's a bug. There is a compile error in a few of the JUnit tests. You can
either change the source in jbpm yourself to remove those errant lines (just
used for debug), or not build those tests by editing the ant build.xml
View the original post :
You could follow the same steps we took to deploy on Websphere, and that was to
experiment.
All you need for jBPM is its .jar and supporting files. We didn't use the
webapp/console, we made our own.
View the original post :
Can you show me your xml how you disabled transactions in jBPM?
I found an error in the documentation, so your transactions might not actually
be disabled.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3965735#3965735
Reply to the post :
Just use a decision handler at P3 to determine if it goes back to P1 or
continues to P4.
The user guide and test cases show how to do this. Either you have a
programmatic handler, or you have a user decision.
View the original post :
You could rely instead of a jBPM timer in general for a
cancellation/reassigning. You could give a user 1 day to do something,
otherwise it's gone.
It's not particularly fair to kill something for a user if his browser closes.
Maybe his computer crashed, his network dropped, accidentally
Ah, I failed to notice you were using the command executor thread.
Yeah, looking at the code, it sure looks like a bug.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3965753#3965753
Reply to the post :
It can be. You have have your product programmatically add your action
handlers to the process definition before you deploy a provided process to your
database.
You have full control over the process definition with the api.
View the original post :
yxyang wrote : (1)what do you mean task end? I went through the source code
and find out that the hibernate queries for findTaskInstance() related
methods only retrieve the open tasks.
This is in the user guide. Task end... means the task is ended.
yxyang wrote : (2)What is the purpose
jbpmndc wrote : What are situations when you should use token.signal instead
of token.end()?
|
| It seems to me you should always use token.end. It's more clear and
intuitive.
There's a difference between token.end and token.signal.
Token.end will END the token and all of its children.
jbpmndc wrote : anonymous wrote : taskInstance.end ends the task, but
SIGNALS the token to the next node.
|
| I understand this. I'm wondering why anyone would use token.signal in
their code. If task.endInstance signals the token to move to the next task,
then writing in your code
Pasting the error would help
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3964354#3964354
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=3964354
___
jboss-user mailing list
I know the user guide mentions using logging for undo capability (IN THE
FUTURE). I wouldn't doubt it was removed for complexity, because man we talked
about this again and again, and it's just crazy.
The point of Graph Oriented Programming... is to follow the graph! If you want
undo paths,
Depends on how you're having the token move. Are you doing a token.signal, or
a taskInstance.end?
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3964356#3964356
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=3964356
What error?
cpob wrote : Pasting the error would help
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3964379#3964379
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=3964379
I really can't follow what the problem is without you showing us the error you
are getting along with the stack trace..
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3964434#3964434
Reply to the post :
Trying to remove things straight from the DB can be hairy, as there are a lot
of interconnected tables.
It's there in the API, look at GraphSession and it's method
.deleteProcessDefinition
You get the graphSession from jbpmContext.getGraphSession()
View the original post :
So, for clarity's sake, you want to have the timer create a new task every n
days? So, it would just sit on the same node and generate tasks every so
often? Or it would progress through the graph?
Implementing timers w/ the included scheduler is very simple. There are JUnit
tests with the
Uh??, yes you will need to write code, how else would you use a library/engine?
There is no other JMX functionality for jBPM, and that startDatabaseManager()
is a Hypersonic JMX call (I think), not a jBPM call.
View the original post :
We potentially talked about having this functionality where we are using jBPM,
and came up with the same idea of drawing transitions back to everywhere.
However, the tasks that have already been performed would still be completed,
unless you programmatically did some magic to undo whatever that
Have you tried running without dba permission?
Off the top of my head, as long as it has rights to select/insert/update/delete
you should be fine.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3963879#3963879
Reply to the post :
Oh, I think that is depending on your sequencer. Doesn't look like you're able
to update the ID, there's gotta be a missing permission somewhere
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3963888#3963888
Reply to the post :
No, the GPD (Graphical Process Designer) is what is responsible for creating
the image.
If you programmatically create your Process Definition, you will have no image,
unless you make your own :(
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3963628#3963628
But with that, Ronald, don't you have to still manipulate the graph in the GPD?
Because if you bring in a prodef xml, all the nodes are going to be glommed
together in the top corner I thought.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3963639#3963639
cwad, did you look at the sample web application? It's in there.
Plus, it is in the manual, although rather light:
Manual wrote : The timer component software is packaged in the core jbpm
library, but it needs to be deployed in one of the following environments:
either you have to configure
Instead of creating multiple task nodes, you can create multiple tasks IN a
task node. So you could then add new tasks programmatically to B, and then
those would all be finished, and then it would move to C.
Only thing is, that would have many B's going to a single C, unless you then
Yes, search the forum, I've answered this 2+ times now.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3963243#3963243
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=3963243
___
I wasn't (at least not intentionally) trying to be rude when I replied. I just
remember posting about it at least twice in the last week or so.
What I usually do, when I post on forums after searching for something with no
results... I mention in my first post my previous search results.
I'm not on the JBoss team, and I don't think kukeltje is either (but I might be
wrong). We are just developers, using jBPM, and trying to give back.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3963298#3963298
Reply to the post :
If you have the schedulerthread running in the background (with a startup
thread) it will load up jBPM and the mappings, etc
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3962830#3962830
Reply to the post :
You could always have a servlet with load on startup, which just does something
simple w/ jBPM, it would load all the classes/mappings/etc at app startup.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3962896#3962896
Reply to the post :
Decision node to two separate forks.
The first fork would go to each of the three tasks. The second fork would go
to only two of them. All three of the tasks would go to the single join.
Something like this
?xml version=1.0 encoding=UTF-8?
|
| process-definition
There are test cases for each of the workflow patterns.
However, some of the code used to 'enable' these patterns aren't really
supported yet in the PDL or Database.
I know the join has a .setNOutofM() (or something like taht) which will allow a
join to only require 2 out of 4 (for example)
I'd suggest just digging through jBPM's source code, tests, documentation, etc.
so you feel more comfortable with jBPM if you are going to be writing your own
nodes.
There are examples of creating action handlers (not on forks, but in general)
in the JUnit tests provided. Plus, the
The JUnit tests also have xml snippets inside them that represent simple/varied
process definitions.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3961959#3961959
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=3961959
Your tier that calls jBPM, could just be a little smarter.
It could detect if it is a task node or a state, and then end/signal
appropriately. Your tier should know if it is a task node or not, and follow
up properly.
View the original post :
Should be rather simple.
You can see how to programmatically create a timer if you look at the
CreateTimerAction source code. They build a Timer object.
I would created an action handler on the node-exit of the start state which
would take the entered date, and then do simple java date math
I'm a little confused. If you say you only use Task Nodes, then why would you
be worried about signalling a state or other node ?
If you signal the tokens, the tasks will never end. It's technically ok if
they are open, though it might be confusing.
Why not just .end() the task, since you
GPD 3.0.11 won't work with Eclipse 3.1.x, upgrade to Eclipse 3.2 if you can, or
use GPD 3.0.9.2 with Eclipse 3.1.x
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3960969#3960969
Reply to the post :
a Gmail notification specifically, or just an email notification?
Currently there is no built in email support in jBPM. You'd have to do it how
you would in 'normal' Java, google around for a bunch of tutorials.
View the original post :
Look at the source code in CreateTimerAction and CancelTimerAction, they create
and cancel timers programmatically.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3961054#3961054
Reply to the post :
I know in 3.1.1 there was the JpdlXmlWriter class, though it is deprecated, and
says xml generation was never finished and will be removed in the future.
If your process is simple it might work.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3960658#3960658
Oops, sorry, I didn't read instance, i just saw process and my mind took
control there :)
Yeah, we wish the JpdlXmlWriter (or something similar) was implemented, because
we are programmatically modifying our process definitions before we deploy it.
It'd be nice to get a new xml back which
I don't have access to commit to the CVS (nor can I do so from work). I
created/commented on a Jira though.
The issue I talk about, is in the action handler. When you have the
executionContext, and do a .getTimer() on it (during a timer action event), it
would always return null.
My fix
I would not suggest jBPM for what you want to do. Just create a simple
jsp/servlet, or use Struts, Spring, Java Server Faces (JSF).
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3960342#3960342
Reply to the post :
biggef wrote : cpob wrote : According to the XML Schema, you can put a
create-timer element on a process definition, but it never did anything through
my testing
| ie nothing happens when the timer fires ?
| Was the timer cancelled when leaving a node ?
It never created the timer
Good point. It's simple enough to change the jBPM code to do what I want, but
I'm just curious as to any though process behind what is done currently.
Also, I wanted to bring up the question if bug JBPM-654 was actually completed,
or marked closed incorrectly?
View the original post :
Create an action handler and register it with the task-create.
Have that handler create a new timer per whatever rules/data you need (which
could be from your database, or an context variable).
Code to create a timer: (This is from the CreateTimerAction code - around line
72)
Timer timer =
http://jira.jboss.com/jira/browse/JBPM-710
Created the JIRA issue, will create new thread in design forum.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3960424#3960424
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=3960424
Why alter the foreign keys? Why not just set the JBPM_TOKEN.PROCESSINSTANCE to
null on the token table, delete the process instances, and then the tokens.
Also, you might want to look into the GraphSession. In 3.1.x, there is a
GraphSession.deleteProcessInstance which will automatically go
:) Yeah, I commented on the existing Jira already.
Sounds like Ronald needs a break ;)
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3959557#3959557
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=3959557
The GPD should have more information about it. There's only the readme.txt in
the .zip which is a basic 'how to install'.
http://www.jboss.com/products/jbpm/downloads Doesn't have any notes for the
GPD. Anything like release notes, or change logs, etc would be great :)
View the original post
I was creating a sample workflow just to use, and I tried having two task-nodes
go to the same fork. Once a fork already has one incoming transition, it will
not allow me to add a second one.
Example: I have a decision node that can go to one of two task-nodes. After
either path, the flow
According to the docs at
http://docs.jboss.com/jbpm/v3/userguide/processmodelling.html#nodetypefork
anonymous wrote : 9.3.5. Nodetype fork
|
| A fork splits one path of execution into multiple concurrent paths of
execution. The default fork behaviour is to create a child token for each
If you look at the ProcessImageServlet, it is getting the image from the
database.
You could do the same thing with all of the files associated with the process
definition.
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3959122#3959122
Reply to the post :
I almost posted this yesterday, but I figured it was obvious, also your other
screen shot showed the class implementing the ActionHandler interface, so I
assumed it still did.
It needs to implement org.jbpm.graph.def.ActionHandler
View the original post :
According to the release notes here
http://sourceforge.net/project/shownotes.php?release_id=433036group_id=70542
[JBPM-593] - suspend() and resume() failing in JBPM 3.1 and
[JBPM-696] - Field Instanciator problems
are changes in 3.1.2.
However, if you look at the JIRA
Since I've been using GPD 3.0.9.2 with Eclipse 3.1.2, I wanted to run GPD
3.0.11 with the same Eclipse version.
I created a new folder with a fresh install of Eclipse 3.1.2 from the zip. I
then downloaded the jbpm-gpd-feature-3.0.11.zip, and extracted it into the
eclipse directory as I did
The jbpm-gpd-feature-3.0.11.zip does not have this issue. I was able to
install the feature zip into Eclipse 3.2 without an issue (3.1.2 doesn't work
though, but that's another thread)
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3959194#3959194
Reply to
Yeah, I just tested a loop of 3 task-nodes (each with a task) and the 3rd goes
back to the first.
New task instances get created for every entry into the task-node (using
default task-node settings, jbpm 3.1.1)
View the original post :
The problem here is that many people get the impression that the sample webapp
is the console. They think it is the way to do it, and it is responsible for
managing jBPM.
It's a SAMPLE webapp, giving EXAMPLES on how to monitor/etc. I think something
needs to be done (no clue what) about
Is a jBPM expert needed for just designing process definitions (workflows)?
Or are you looking for someone who understands the jBPM API, the nuances with
the GPD plugin, the jPDL xml schema itself, the database structure, etc?
If I was close by, I'd be banging on your door :)
View the original
Yes, I second the sticky thread, something like About the jBPM Sample Webapp
or something similar. :)
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3959253#3959253
Reply to the post :
http://www.jboss.com/index.html?module=bbop=postingmode=replyp=3959253
Koen's Announcement wrote : Both GPD releases are build and tested against
Eclipse 3.2 and WTP 1.5.
Says nothing about 3.0.10 for 3.1.x, anyway, I tried 3.0.10 for Eclipse 3.1.2
and it still didn't work :(
View the original post :
It worked for you on Eclipse 3.1.1 w/ the 3.0.11.feature zip?
Hmm, I wonder if eclipse 3.1.1 vs 3.1.2 could have differences?
View the original post :
http://www.jboss.com/index.html?module=bbop=viewtopicp=3959316#3959316
Reply to the post :
1 - 100 of 111 matches
Mail list logo
